Female and Male Admission to Graduate School: An Illustrative Inquiry

Female and male admission rates to four graduate programs at the University of North Carolina at Chapel Hill are compared for 1972–73 and 1973–74. To assess possible sex-related bias in admission, rates are adjusted for applicant qualifications by analysis of covariance and by balancing. Guidelines for choosing between these two techniques are developed. The use of adjustment techniques for these data is shown to be justified in terms of Meehl’s analysis of the ex post facto design. Other adjustment procedures, direct and indirect standardization, are critically discussed. The adjusted admission rates reflect, in one case, i.e., for one program and one admission year, a slight advantage for male applicants over females, while in three cases, female applicants were granted a slight advantage over males in admission. In the remaining four cases, there is no evidence that sex of applicant, per se, played a role in admission decisions. Wherever a sex-related advantage is detected, the favored sex is that with the fewer applicants to the program. While the generalization of these results to other programs and other graduate schools is unwarranted, it is appropriate to emphasize the value of applying balancing and/or analysis of covariance to data that may be collected for the purpose of detecting possible discrimination in the selection or evaluation of identifiable population subgroups.